Download 11 Sexual segregation in ungulates: from individual

P1: FYX
0521835224c11.xml
CUUK074B-Ruckstuhl
September 5, 2005
18:4
richard bon, jean-louis deneubourg, jean-françois
g e r a r d a n d pa b l o m i c h e l e n a
11
Sexual segregation in ungulates:
from individual mechanisms to
collective patterns
overview
Sexual segregation is an integral aspect of the socio-spatial organization
of ungulate populations. Very often, the social, spatial and ecological
components have been confounded (Bon, 1992) and we have argued that
it is necessary to define and distinguish between each of them (Bon &
Campan, 1996; see also Chapter 2 by Larissa Conradt). In the present
chapter, we point out that sexual segregation is a complex phenomenon
that can be produced by distinct mechanisms. One of the main issues is
to know whether segregation by habitats necessarily derives from sexual
difference in habitat choice, or can derive from alternative causes, i.e.
spatial and social mechanisms (see also Chapter 2). Habitat segregation
implies heterogeneous habitat (Miquelle et al., 1992), which we assume
not to be obligatory for social and spatial segregation to occur. We
distinguish hypothetical mechanisms relevant only in a heterogeneous
environment from those relevant in both heterogeneous and homogeneous environments. We focus on behavioural mechanisms that may
generate social and spatial segregation/aggregation, and the problem
of the scale at which segregation may occur. Finally, we suggest that
segregation cannot only be considered as a result of individuals behaving independently of each another, but also as a result of interactions
between individuals on a larger (population) scale.
Habitat versus social segregation
Miquelle et al. (1992) noted that differences in habitat selection often
lead to sexual segregation and resource partitioning between the sexes.
Sexual Segregation in Vertebrates: Ecology of the Two Sexes, eds. K. E. Ruckstuhl and P. Neuhaus.
C Cambridge University Press 2005.
Published by Cambridge University Press. 180
P1: FYX
0521835224c11.xml
CUUK074B-Ruckstuhl
September 5, 2005
18:4
From individual mechanisms to collective patterns
For many researchers, sexual segregation is the differential use of space
by the sexes (Kie & Bowyer, 1999; Weckerly et al., 2001). This statement is
motivated by the conviction that the sexes are segregated because of differences in habitat choice (see Habitat segregation, Chapter 2). However,
social and habitat segregation are not necessarily linked: segregation
can occur by using the same resources but at different times (Francisci
et al., 1985; Jakimchuk et al., 1987) or by using different resources within
the same areas (Staines et al., 1982; Bowyer, 1984) (see Chapter 2 and
3). In addition, it is difficult to decide whether segregation derives
from habitat choices or is a consequence of differences in social or spatial behaviour in the wild (Shank, 1982; LaGory et al., 1991). The idea
that sexual segregation is strictly determined by differences in habitat choice can be challenged by a point of semantic and by empirical
evidence.
Understanding sexual segregation implies defining its meaning
(Bon & Campan, 1996; Main et al., 1996). Segregation comes from the
Latin segregare ‘separate from the flock, isolate, divide’ or the Greek
σ ε ‘apart from’ and γρεξ ‘herd, flock’ which means separating, isolating an individual or a group from conspecifics (Chambers Encyclopedic
English Dictionary, 1994; Grand Usuel Larousse, 1997). Thus, etymologically
speaking, segregation refers to a socially motivated action, although
this social component has been considered secondary by most authors.
The phrase ‘males and females live apart’ can mean living in distinct
groups, living in distinct areas, or living in distinct habitat types (Bon,
1998; Bon et al., 2001). While sexual segregation is most often considered as ecologically determined (Polis, 1984), we argue that it is relevant
to recognize the social, spatial and habitat components/dimensions
that sexual segregation may involve as well (Main & Coblentz, 1990;
Bon, 1991; Weckerly, 1993; Miquelle et al., 1992; Bon & Campan, 1996;
Conradt, 1999).
Sexual segregation is an outcome at a population level, resulting
from several possible mechanisms (Fig. 11.1). It is therefore necessary to
define the components of sexual segregation as objectively as possible,
and without inferring from the supposed individual mechanisms (see
Chapter 2). In this chapter we will mainly be concerned with social
segregation and the mechanisms supposed to be involved in it.
We propose to define social segregation as the trend for individual
animals to aggregate with animals or subjects belonging to the same
social category, e.g. sex and age (Bon & Campan, 1996; see also Conradt,
1998b). Before developing hypotheses involving mechanisms that may
generate social and spatial segregation, we state some basic conditions
to consider gregariousness.
181
P1: FYX
0521835224c11.xml
182
CUUK074B-Ruckstuhl
September 5, 2005
18:4
Richard Bon et al.
Habitat A
Habitat B
Habitat C
Habitat D
Habitat C
Figure 11.1 Theoretical population of dimorphic and social ungulates
in which sexual segregation involves high levels of habitat, spatial
and social segregation. (1) Habitat segregation: habitat A is only used
by males, while habitat B is only used by females. (2) Spatial
segregation: the females are located near the centre and the males
at the periphery of the area occupied by the whole population;
furthermore, habitat C is used by females or males according to the
location of the corresponding patches. (3) Social segregation: singlesex groups are more frequent than expected by chance and this
remains true within a portion of space supporting a single habitat
and used by both sexes (habitat D patch).
It is important to recognize that animals could aggregate simply
as a result of individuals of solitary species being attracted by the same
environmental stimulus (feeding patches, refuge or migration corridors, for instance) without any social attraction. However, when the
attractive environmental stimulus disappears, the groups will dissolve.
Accordingly, all hypotheses discussed in this chapter implicitly assume
that the species concerned are social, i.e. individual animals aggregate
in more or less stable groups.
Some assumptions must be met for animals to aggregate in
groups (see also Krause & Ruxton, 2002). Aggregation occurs by interattraction between mobile individuals, via visual contact for most of the
wild ungulate species (see Gerard et al., 2002), even though olfactory
or auditory stimuli may also be involved (Barrette, 1991). In addition,
interactions between individuals are necessary to keep group cohesiveness, which implies co-ordinated activities and thus allelo-mimetism
(Deneubourg & Goss, 1989). See Box 11.1.
Animals are classically considered to associate at random (Grubb
& Jewell, 1966; Geist, 1971; Langman, 1977; Hillman, 1987; Hinch et al.,
1990), with grouping depending on food distribution (Lott & Minta,
1983; Lawrence, 1990). However, ecological factors alone cannot account
for phenotype assortment according to body size, sex, age or social
status (Estes, 1991b; Bon et al., 1993; Villaret & Bon, 1995, 1998; Cransac
et al., 1998). Social segregation between sexes is a particular case of
social aggregation, as two categories of individuals are found together
less often than expected if they were associated at random (Conradt,
1998b). Concerning the mechanism involved, this means that the sexes
183
– Discrimination-based fusion
(involving attraction/repulsion)
depending on:
– social compatibility based on previous
encounters
– assortive phenotype (sex, age, size)
based on activity compatibility
(movement, activity budget)
– co-ordination in movement
– synchrony in activity budgets
– Long-lasting or frequent interactions
– Symetrical interactions
– Allelo-mimetism
September 5, 2005
– No discrimination between
individuals
Social mechanisms conditioning
group cohesiveness
CUUK074B-Ruckstuhl
Social mechanisms conditioning group formations
0521835224c11.xml
Box 11.1 Social mechanisms conditioning group formation and cohesiveness
P1: FYX
18:4
P1: FYX
0521835224c11.xml
184
CUUK074B-Ruckstuhl
September 5, 2005
18:4
Richard Bon et al.
(social classes) can differ in the degree of social attraction to the opposite sex, move at a velocity or/and have activity rhythms that impair
association for long periods of time (see also Chapter 10). Social attraction can rely on the activity of animals but also on the capacity to
discriminate the sex or age of conspecifics. Experiments have revealed
that domestic ungulates were capable of social discrimination between
juveniles and adults (Kendrick et al., 1995; Porter et al., 2001). The sexes
can be segregated on a large scale, involving high degrees of spatial
segregation, or on a small scale. On a small scale, subgroups can
be detected within larger groups as occurs between juveniles within
female groups (Richard & Pépin, 1990; Gerard et al., 1995). In children,
girls and boys may be socially segregated at a smaller scale by being
closer to their own gender than expected by chance in a school playground (see Pellegrini et al., 2003; Chapter 12). The local interactions
between neighbours generating this small-scale segregation include
attraction and repulsion as well as mutual adjustment of activities.
Furthermore, the sorting process may be facilitated by simple physical
constraints such as crowding of one sex that reduces available space
for the other sex. These rules can lead to group splitting and so the
promotion of social segregation at a larger scale (see Deneubourg et al.,
1991 for an explanation in social insects).
In the following sections, we present some habitat-related mechanisms leading to sexual segregation in heterogeneous environments.
We then forward arguments that illustrate the importance of nonecological mechanisms leading to social and/or spatial segregation at
large and small scales in both heterogeneous and homogeneous environment. Lastly, we suggest that unexplored processes such as ‘social
amplification’ can produce higher level of habitat or spatial segregation
that cannot be obtained if individuals behaved independently of one
another.
Habitat choice, parental behaviour, social disturbances
and predation risks
In ungulates, males are not implicated in raising young. The reproductive strategy or predation risk hypothesis states that sexual segregation is the consequence of sexual difference in reproductive investment (Chapters 3 and 9). Females’ use of safe habitat is considered
as an adaptation to reduce the risk of predation on offspring and as
a way to improve females’ reproductive success. However, only a few
authors have considered the mechanisms promoting changes of female
P1: FYX
0521835224c11.xml
CUUK074B-Ruckstuhl
September 5, 2005
18:4
From individual mechanisms to collective patterns
behaviour around parturition. The following examples will illustrate
the proximate explanation of social and habitat segregation.
In mountain ungulates, females restrict themselves to steep slopes
just before parturition, meanwhile non-parturient females and males
use better feeding habitats (Shank, 1982; Bergerud et al., 1984). Spatial
segregation is less marked when involving females without offspring
in mountain and non-mountain areas (Festa-Bianchet, 1988; Miquelle
et al., 1992; Bon et al., 1995; Ginnett & Demment, 1999). But the choice of
habitat type depends on the local context, including prey and predation
types (Ruckstuhl & Neuhaus, 2002). For instance, females with calves in
Masai giraffes, Giraffa camelopardalis tippelskirchi, and kudu, Tragelaphus
strepsiceros, avoid woodland habitats and use open habitats probably
because higher visibility provides better risk detection (Du Toit, 1995;
Ginnett & Demment, 1999). The choice of secure areas around parturition suggests that females are more sensitive to disturbance and
predation risks at this period (Frid, 1999; Weckerly et al., 2001). In areas
where large predators are scarce or absent, sexual segregation in habitat use vanishes (du Toit, 1995). Berger et al. (2001) showed that naı̈ve
female moose, Alces alces, experiencing predation on their calves were
able very quickly to exhibit anti-predator behaviour. Kohlmann et al.
(1996) reported that female Nubian ibex, Capra ibex nubiana, with young
kids temporarily confined in a predator safe canyon, differ in habitat
use from females followed by kids. The former move farther from escape
terrain, use better feeding habitat and spent more time feeding than
the latter.
However, there is also a much more proximal explanation why
females search out particular areas. Increased habitat and spatial segregation is also the consequence of marked modification of parturient
females’ social behaviour (Poindron et al., 1988). A few days before parturition or when followed by neonates, females become asocial (Alexander
et al., 1979; du Toit, 1995) and aggressive (Gosling, 1969; Cederlund,
1987; Estep et al., 1993) in several ungulate species. By withdrawing
from groups, female taruca, Hippocamelus antisensis, near parturition segregated by habitat from male and mixed-sex groups commonly found
year round (Merkt in Frid, 1999). Miquelle et al. (1992) reported that
female moose with offspring seemed to avoid areas already used by
other moose. Cliffs and forested areas provide physical obstacles disrupting visual contact and so can be chosen by parturient females as
they allow seclusion from conspecifics of both sexes (Cransac et al., 1998)
and animals of other species, including men and predators. Seclusion
would facilitate the mother--young bonding (Poindron et al., 1988) and
185
P1: FYX
0521835224c11.xml
186
CUUK074B-Ruckstuhl
September 5, 2005
18:4
Richard Bon et al.
allow avoiding social perturbations or the possibility of adoption by
other parturient females (Arnold et al., 1975). In addition, other physiographic characteristics can be key factors involved in the selection of
areas to give birth (see Bon et al., 1995).
Social and habitat segregation is supposed to be determined by
gestation and the presence of young. Thus, segregation might peak
during the birth season. Behavioural changes around parturition, associability and aggressiveness, are not caused by habitat heterogeneity,
although females can use it at that period to satisfy isolation, as
discussed earlier. Thus, we can expect social and spatial segregation
between parturient, lactating and non-lactating females or males in
homogeneous environments. Because the maternal behaviour is hormonally induced, social and habitat segregation should vanish with
the end of maternal care and the physiological weaning. If it persists
outside the period of maternal care, it is necessary to consider other
mechanisms than those invoked by the predation risks hypothesis. It
remains also to be explained what causes females to venture farther
from safe areas when offspring become older (Bergerud et al., 1984;
Bon et al., 1995).
is habitat segreg ation equal to sexual
differences in habitat choice?
Up to now, with a few exceptions, sexual segregation was mostly
attributed to different habitat selection by the two sexes (see Bon &
Campan, 1996). However, is habitat segregation necessarily caused by
different habitat choices between the sexes? Some authors argued that
habitat segregation might result from other mechanisms, such as, for
example, social mechanisms (Shank, 1982; LaGory et al., 1991; Bon,
1991).
Social mechanisms involved in segregation
Social segregation versus habitat segregation
Conradt (1998b, Chapter 2) proposed an index that allows measuring
and comparing the degree of social, spatial and habitat segregation
between the sexes. Social segregation here refers to the presence of
males and females in single-sex groups, while spatial segregation refers
to the use of exclusive quadrates by one sex. Using long-term studies on
red deer, Cervus elaphus, and Soay sheep, Ovis aries, Conradt (1999) showed
that the degree of social and spatial segregation was always higher than
P1: FYX
0521835224c11.xml
CUUK074B-Ruckstuhl
September 5, 2005
18:4
From individual mechanisms to collective patterns
habitat segregation. This allowed the author to conclude that at least
one part of social segregation cannot result from habitat segregation,
and that each component was probably the result of different causes
(Bon & Campan, 1996). Social segregation seems to be a rule in social
dimorphic ungulates, and independent of the size of populations (Bon
et al., 2001), density of males or females (Conradt et al., 1999b) and
spatial segregation (Kie & Bowyer, 1999).
After a control of predators in an enclosed population of whitetailed deer, Odocoileus virginianus, Kie and Bowyer (1999) reported that
spatial segregation decreased at high density of deer, whereas the level
of social segregation was unchanged. As a consequence of higher spatial overlap between the sexes, dietary differences were lower than at
moderate density and diet impoverished more in females than males.
This is inconsistent with a prediction of the scramble competition
hypothesis (see Chapter 2) that females actively or passively exclude
adult males from preferred areas (Bleich et al., 1997; Romeo et al.,
1997). In their study, Kie and Bowyer (1999) also rejected the social
factors hypothesis, namely, that social segregation was driven from
males avoiding costly social interactions linked to female proximity
in mixed-sex groups. This hypothesis has been criticized because sexual interactions are dependent on sexual hormones that are produced
seasonally, and so it is unlikely to apply outside the mating period (Main
et al., 1996). However, this does not exclude the relevance of other social
mechanisms.
Social segregation based on age or social status
More rarely considered, age is a factor that is implied in the degree
of sexual segregation (Bon et al., 1993; Bon & Campan, 1996). Yearling
males are most often observed in female groups while the oldest ones
are rarely associated with females outside the rut (Nievergelt, 1967;
Geist, 1971; Bon & Campan, 1989; Festa-Bianchet, 1991; Miquelle et al.,
1992; Ruckstuhl, 1998; Ruckstuhl & Festa-Bianchet, 2001). Bon et al.
(2001) found a gradient of social segregation linked to male age in
Alpine ibex, Capra ibex, even when spatial segregation was low in winter, rendering ecological mechanisms a very unlikely cause of social
segregation. Age difference was also found to be an important factor of
social and spatial segregation among males splitting up into groups of
similar-aged individuals (Bon et al., 1993; Villaret & Bon, 1995, 1998).
Social segregation within the sexes is not only observed as a function of age, but may also be dependent on events occurring early in
187
P1: FYX
0521835224c11.xml
188
CUUK074B-Ruckstuhl
September 5, 2005
18:4
Richard Bon et al.
ontogeny. Jewell (1986) reported that castrated Soay sheep males not
only formed groups of their own but also used distinct home ranges.
They socially and spatially segregated from entire males and females
(see also Ruckstuhl et al., submitted). It is likely that the lack of male
hormones that is implicated in the male-like behaviour influenced the
nature of interactions and levels of behaviour of early castrated males.
As a consequence, these males set up a social network among themselves, which made them socially segregated from non-castrated males
and females.
The assumption that aggregation is based on a general interattraction between conspecifics must be modulated because the force of
attraction may vary during ontogeny. When ageing, individuals seem
to be less sociable in some populations of European mouflons (Ovis
aries), bighorn sheep, chamois (Rupicapra rupicapra) and isard (R. pyrenaica) (Pfeffer, 1967; Geist, 1971; Shank, 1985; Richard-Hansen, 1992; Hass
& Jenni, 1993). According to Shank (1985: 122), social and spatial segregation of old male chamois is a trade-off between dependence on
feeding resources and a ‘need for solitude’, reflecting social intolerance.
However, old animals might more often be alone because they are less
sociable or because they lack similar-aged peers (Villaret & Bon, 1998).
When available, old non-reproductive buffaloes were reported to group
together, apart from younger males (Sinclair, 1977). Population density
and hence the probability of meeting conspecifics is also a factor contributing to the chance of both sexes to be found alone (Fig. 11.2; see
also Gerard & Loisel, 1995 and Gerard et al., 2002 for a discussion on
mechanisms underlying aggregation).
b e h av i o u r a l m e c h a n i s m s i n d e p e n d e n t o f
habitat heterogeneit y
Recently, new hypotheses were proposed, suggesting that sexual segregation may be explained by different mechanisms, including movement or spatial behaviour, activity budgets, and social behaviour (Bon &
Campan, 1996; Conradt, 1998b; Ruckstuhl, 1998; Bon et al., 2001). These
hypotheses differ notably from previous ones in the sense that they do
not depend upon habitat heterogeneity, and that they propose mechanisms that can produce social and eventually spatial segregation in
heterogeneous but also in homogeneous environment. If social segregation is observed in controlled and homogeneous habitats, observers
have either not detected an ecological heterogeneity that animals do
detect, or non-ecological mechanisms are at work.
P1: FYX
CUUK074B-Ruckstuhl
September 5, 2005
18:4
From individual mechanisms to collective patterns
20
Bargy
Sous-Dine
18
16
14
% alone
0521835224c11.xml
12
10
8
6
4
2
0
Females
Males 1
Males 2–3 Males 4 –5 Males 6 –8 Males = 9
Figure 11.2 Proportion of observations corresponding to alone
individuals during census of groups in two populations of Alpine
ibex, Bargy and Sous-Dine, according to the sex and male age. Bargy
population contained c. 120 adults while Sous-Dine population only
27 adults. Data from the mating and the birth periods were omitted
because males are often alone during the rut and females less social
at parturition. Males were found more often alone when getting older
in both populations. Note, however, that the probability of being
found alone in both sexes is higher in the smaller than the larger
population. Numbers in brackets correspond to the total number of
observations of groups (lone individuals included).
The activity budget hypothesis
Recently, it has been proposed that sexual dimorphism in body size
could lead to sexual differences in activity budgets in ungulates
(Conradt, 1998a; Ruckstuhl, 1998; Chapters 2 and 10). One basic prediction of the activity budget hypothesis is that, because of their smaller
size, higher energy requirements and lower efficiency in processing forage, females would spend more time feeding than males (Ruckstuhl,
1998; Ruckstuhl & Neuhaus, 2000; see also Ruckstuhl & Neuhaus
Chapter 10). The resulting asynchrony in activity is considered a major
constraint for mixed-sex groups to be maintained, resulting in social
segregation.
The activity budget hypothesis predicts segregation without
implying differences in forage selection. It is theoretically possible to
find social segregation without habitat/spatial segregation: both sexes
can form separate groups, use overlapping ranges in which they can
exploit the same habitat patches at different times (Francisci et al., 1985)
or at the same time without mixing. The activity budget hypothesis
189
P1: FYX
0521835224c11.xml
190
CUUK074B-Ruckstuhl
September 5, 2005
18:4
Richard Bon et al.
assumes that to stay together, the individuals belonging to the same
groups must share similar activity rhythms allowing activity synchrony.
Allelo-mimetism is implicit to activity synchrony, i.e. when individuals
in a group exhibit patterns of individual activity that would not occur
if individuals were independent (Deneubourg & Goss, 1989). The hypothetical possibilities of animals meeting and having similar activity
rhythms are illustrated in Fig. 11.3(a) and (b). Consider a population
of individuals with two states, active and inactive, and independent
from each other in their activity. A stable group will depend upon
the probability of two individuals to be synchronized in order to stay
together. If the activity budgets differ too much between both animals (as suggested to occur between females and males in dimorphic
species), the probability of staying for a time longer than a bout of
activity or inactivity is unlikely (Fig. 11.3(a)). If two animals having the
same activity rhythm meet, the probability of associating for a lasting period will depend on both individuals being in the same phase
(Fig. 11.3(b)). This probability would be higher for same-sex than for
opposite-sex animals. However, synchrony in activity between the sexes
is possible if males and females do not vary too much in activity budgets, and if at least one sex adjusts its activity rhythm to that of the
other sex. Although we do not know how overall activity synchrony
in a group is achieved, it can be assumed that having the same activity as surrounding animals can result in a high degree of overall synchrony. It is not necessary for individuals to adjust their behaviour
to the entire group. In most ungulate species, groups are unstable
in size and composition (Marchal et al., 1998). However, data obtained
from wild populations indicate that activity synchrony in single-sex
groups is higher than expected by chance (Côté et al., 1997; Ruckstuhl,
1999). This suggests that animals belonging to groups of the same sex
are either synchronized by the same external releaser or possess the
same internal clock. It is more parsimonious thinking that individuals with activity budgets not too different can tune their activity to
each other through interactions (Fig. 11.3(c)), such as allelo-mimetism
allowing to be in phase. Ramı́rez Ávila et al. (2003) have shown that
interactions between oscillators that differ in their intrinsic period
enable individuals to adopt the same period. This individual ability
could lead to clustering of individuals having similar activity periods,
and social segregation between individuals having dissimilar periods.
Note that the possibility for individuals to aggregate based on similar activity period and interaction does not mean that all groups are
synchronized.
P1: FYX
0521835224c11.xml
CUUK074B-Ruckstuhl
September 5, 2005
18:4
(a) Individual animals with different rhythms
Active
Inactive
Lack of synchrony
(b) Individuals with the same but unsynchronized rhythm
Active
Inactive
Lack of synchrony
(c) Individuals with similar rhythm, and which can
synchronize activity to each other through interaction
Active
Inactive
Possibility of
activity synchrony
High level
of synchrony
Figure 11.3 Hypothetical cases of meeting between two individuals
and possibility of activity synchrony: (a) both individuals exhibit
distinct or (b) similar activity rhythms without possibility of
synchronization and (c) both individuals can synchronize their
activity.
191
P1: FYX
0521835224c11.xml
192
CUUK074B-Ruckstuhl
September 5, 2005
18:4
Richard Bon et al.
Dispersal behaviour
The impact of sexual differences in dispersal behaviour on social and
spatial segregation has rarely been examined (Bon & Campan, 1996).
In dimorphic ungulate species, as in many mammals, juvenile females
usually settle in or near their maternal or natal range while males
often disperse from their natal group or area (cervids: Clutton-Brock
et al., 1982; Bunnell & Harestad, 1983; Nelson & Mech, 1984; Cederlund
et al., 1987; Cederlund & Sand, 1992; Hölzenbein & Marchinton, 1992;
wild sheep: Festa-Bianchet, 1986; Dubois et al., 1994). Based upon longterm radio-tracking of mouflons, Dubois et al. (1993) found two categories of two- to three-year-old males regarding their dispersion outside
the rut. Some males still used their natal range, while most of them
gradually dispersed until using non-maternal and stable ranges. Even
when males and females spatially overlapped, they were still socially
segregated (Dubois et al., 1993). The difference in spatial dispersion
between males and females often results in males using a higher diversity of habitats (Ordway & Krausman, 1986; Villaret et al., 1997).
Motion behaviour
Dimorphism in body size can also be at the origin of other differences
in behaviour, such as motion. Larger or powerful individuals probably
walk or move more rapidly than smaller individuals, or travel a larger
distance per unit time. If so, group splitting and spatial segregation,
or structuring at a small scale, can arise even if the individuals in the
population, whatever their size, do not differ in their habitat choices
(Gueron et al., 1996; Couzin & Krause, 2003). Ruckstuhl (1998) found that
bighorn ewes and rams had the same step rate per unit time, but ewes
dedicated more time to walking, had longer walking bouts and travelled
larger distances than rams. Miquelle et al. (1992) also reported that in
moose females and subadult males moved more during the feeding
periods than large males. On the other hand, Michelena et al. (2004)
have shown experimentally that merino rams, when placed within the
same pastures were walking twice as rapidly than ewes within the same
pastures. However, no social segregation at a large scale was found since
both sexes were together in a single group for several weeks. During
the eight-week experiment, males were more often found in the front
of the group than females. Same-sex pairs of nearest neighbours were
significantly more frequent than mixed-sex ones, which might result
from higher step rates in males than in females. However, when sheep
P1: FYX
0521835224c11.xml
CUUK074B-Ruckstuhl
September 5, 2005
18:4
From individual mechanisms to collective patterns
of both sexes were distributed at random within the group, pairs of
nearest neighbours of same sex still outnumbered opposite-sex pairs.
These results suggest a high level of inter-sex attraction, explaining
the lack of social segregation at a large scale, compensating for the
difference in activity budget and moving velocity between the sexes,
and higher intra-sex than inter-sex affinity accounting for the social
segregation at a small scale.
The social affinity hypothesis
Mechanisms proposed to explain social segregation in children, and
data collected on behavioural development and social interactions in
other mammal species inspired the social affinity hypothesis (Bon &
Campan, 1996; Bon et al., 2001). Jacklin and Maccoby (1978) suggested
that the differences in the level of activity (Eaton & Enns, 1986) and in
how both sexes get socially involved could lead to problems of social
matching between girls and boys (see Pellegrini & Long, 2003 for a
recent discussion). They proposed the notion of behavioural incompatibility that Legault and Strayer (1991) extended by defining it as ‘a set
of differences in the overall composition of behavioural repertoire’ to
account for social segregation between the sexes in children. Bon and
Campan (1996) argued that sexual differences in behaviour and social
motivation lead to behavioural and social incompatibility, and thus
social segregation between different sex--age classes in social ungulates.
Behavioural compatibility would be necessary for social cohesion to
occur.
In mammals, sexual differences are found in levels of motor
activity (Holekamp & Sherman, 1989) and type of behaviour (Cheney,
1978; Sachs & Harris, 1978; Moore, 1985; Meaney, 1988) from an early
stage of life. In dimorphic ungulates, social behaviour and morphology
mature more gradually in males than in females, long after reaching
sexual maturity (Geist, 1968, 1971; Grubb, 1974; Jarman, 1983; Rothstein
& Griswold, 1991; Shackleton, 1991). Juvenile males are more often
engaged in rough-and-tumble or pseudo-sexual plays, while females are
more often engaged in locomotor play and also spend more time in
feeding activities than males (Bon & Campan, 1996). The difference
between the sexes in the amount of interactions still persists into
adulthood (Le Pendu et al., 2000). Owing to the differences of social
motivation, behavioural style and morphology, females could avoid or
be indifferent to male social interactions. Although mixed-sex groups
of mouflons were infrequent, Le Pendu et al. (2000) found a high rate
193
P1: FYX
0521835224c11.xml
194
CUUK074B-Ruckstuhl
September 5, 2005
18:4
Richard Bon et al.
of inter-sex interactions, initiated by males over two years old when
males and females co-occurred at attractive feeding sites. Males interact much more frequently than females in sheep, even when the latter are involved in the interactions (Michelena et al., 2004). Several
authors have argued that females avoid interacting with dominant
males (see Bon & Campan, 1996). Weckerly et al. (2001) found that
Roosevelt elk (Cervus elaphus roosevelti) females displayed slightly higher
aggression rates in mixed-sex groups when males were more prevalent,
possibly as a consequence of females approaching one another when
avoiding males. Female and mixed-sex groups also walked away when
approached by male groups exceeding six individuals. Because females
avoided only large male groups, Weckerly et al. (2001) concluded that
this social mechanism is unlikely to account for high degrees of social
segregation.
From a physiological point of view, behavioural dimorphism in
social behaviour and dispersal between the sexes is induced by perinatal androgens (Hinde, 1974; Goldfoot et al., 1984; Moore, 1985; Meaney
et al., 1985; Holekamp & Sherman, 1989). For example, Jewell (1986, 1997)
showed how castrated Soay lambs formed self-contained groups, avoided
interacting with other sheep and used ranges distinct from ewes and
rams as adults (see also Clutton-Brock et al., 1982). These results indicate
how physiological mechanisms and the type of behavioural style can
affect social and spatial segregation. The social affinity hypothesis thus
predicts that grouping will probably occur between animals of the same
sex and age (Bon et al., 2001). Yet, even if groups persist when individuals
share the same motivation to associate, they may contain individuals
with very dissimilar behaviour such as females and offspring. This is
made possible because of the shared motivation to stay together and
because individuals can carry out their maintenance activities within
such groups.
s ynerg y between different mechanisms
All populations occurring in the wild face a certain degree of heterogeneity in their habitat. How an individual animal chooses its home
range will depend on some basic or vital requirements, but also on
phenotypic constraints or cognitive abilities.
Habitat segregation is most often considered as the result of an
active choice or compromising between conflicting factors. However, it
is worth noting that an experimental design is necessary to ascertain
P1: FYX
0521835224c11.xml
CUUK074B-Ruckstuhl
September 5, 2005
18:4
From individual mechanisms to collective patterns
that females and males differ in habitat choice. For instance, Morton
(1990) demonstrated experimentally that habitat segregation observed
in the hooded warbler (Wilsonia citrina) was founded on sexually distinct
preferences of physical characteristics of habitat. However, Desrochers
(1989) showed that male and female black-capped chickadees were segregated by habitat because males excluded females from preferred
microhabitats. Such demonstrations remain scarce for ungulates. PérezBarberı́a and Gordon (1999) carried out an experiment with Soay sheep
and showed that both sexes preferred high quality to low quality grazing patches. However, contrary to the predictions of the forage selection
hypothesis, females spent more time foraging on the low quality swards
than males.
The activity budget and the social affinity hypotheses do not
exclude the contribution of mechanisms linked to reproduction or ecological factors to sexual segregation, but they state that differences in
behaviour and social motivation are basic mechanisms of social segregation. Food quality and distribution, predation risks (Jarman, 1974) but
also population density and habitat openness are causal factors of animal grouping (Barrette, 1991; Gerard et al., 1995, 2002). In the wild, it is
difficult to set apart the impact of ecological factors from that of social
factors and we argue that sexual segregation probably involves several
mechanisms. The question of synergy or antagonism between different
mechanisms, in particular social and ecological ones, is poorly documented (Bon & Campan, 1996). To illustrate the importance of this topic,
we present a model where slight differences in habitat use between the
sexes can be amplified by social attraction (see Appendix 11.1 for details
of the model).
Finally, we would like to point out that experimental studies are
needed to test non-ecological mechanisms in controlled habitat, with
the underlying idea that a better knowledge of behavioural/cognitive
mechanisms and interactions between individuals will provide new
insight into aggregation and segregation dynamics. We also recommend
considering the quantitative aspect and interplay of mechanisms, and
the dynamics of social and spatial structures that are difficult to tackle
if one only considers the individual’s perspective.
a p p e n d i x 11. 1
Let us consider a population of solitary animals (there are no interactions between individuals whatever their sex), composed of two
195
P1: FYX
0521835224c11.xml
196
CUUK074B-Ruckstuhl
September 5, 2005
18:4
Richard Bon et al.
Table 11.1 Probability of individual males or females moving between
habitats A and B, depending on whether they are solitary or social.
Solitary
Males
Females
Social
Probability of
moving from
A to B
Probability of
moving from
B to A
Probability of
moving from
A to B
Probability of
moving from
B to A
αM
αF
βM
βF
α M /(1 + M A )∗
α F /(1 + F A )∗
β M /(1 + M B )∗
β F /(1 + F B )∗
∗
MA and MB (FA and FB ) are respectively the total numbers of males (females) in
habitats A and B. M = M A + MB and F = FA + FB are the total subpopulations of
males and females.
subpopulations: males and females. The individuals can travel freely
between two contiguous habitats A and B. FA and FB (MA and MB )
are respectively the total numbers of females (males) in habitats A
and B. F = FA + FB (M = MA + MB ) is the total subpopulation of females
(males).
Solitary individuals
At each time-step, any female on habitat A or B has the probabilities
α f and β f to move from A to B and from B to A, depending on the
characteristics of habitats A and B (Table 11.1). Accordingly, the number
of females moving from A to B is α F F A and from B to A is β F F B . At the
equilibrium, the number of individual females moving from A to B
equals that moving from B to A. This can be written:
αF F A = βF F B
or
α F F A = β F (F − F A )
It is then easy to find that:
FA =
βF F
(α F + β F )
If f A is the proportion of females in A (F A /F) and f B their proportion in
B (F B /F) and we define rF = α F /β F , then:
fA =
1
(r F + 1)
and
fB = 1 − fA
Similarly the fraction of males in habitat A is:
mA =
1
(r M + 1)
and
mB = 1 − mA
P1: FYX
0521835224c11.xml
CUUK074B-Ruckstuhl
September 5, 2005
18:4
From individual mechanisms to collective patterns
Fraction in habitats A and B
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49
Population size
Figure 11.4 Theoretical fractions of animals in two distinct habitats,
A (circles) and B (squares) as a function of population size F. White
symbols correspond to solitary individuals, and black symbols to
social individuals. For example, if solitary females have a small
individual preference for habitat A, the proportion of females f A in
habitat A will remain to 0.54 and in habitat B f B to 0.46. If males’
preference for habitat B equals females’ preference for habitat A, then
the proportion of males will be such that m B = f A and m A = f B . In
the case of social species, with the same individual preference for
habitat A than solitary animals, the proportion of females (males) will
however increase with the female (male) population size. This figure
shows how social interaction may amplify the initial individual
habitat choice and, in the case of different individual habitat choice
between the sexes, how the degree of habitat segregation increases
with the population size of females and males.
Whatever the size of the female subpopulation, the proportion of
females in habitats A and B at the equilibrium will depend on the
relative preference of females for habitats A (α F ) and B (β F ) and so on
the value taken by rF . If there is no preference for any habitat in both
sexes, then rF = rM = 1, and there are proportionately as many females
and males in both habitats ( fA = fB = 0.5 and mA = mB = 0.5) leading
to a lack of habitat segregation between the sexes. If females prefer the
habitat A and males prefer the habitat B, then α F < β F , α M > β M , f A > m A
and f B < m B . For example, consider the initial population is composed
of individual females with a ratio r F = 0.86, leading to a slight preference for A, then f A = 0.54 and fB = 0.46 (Fig. 11.4). The greater the
difference between the ratio r F and r M , the greater will be the habitat
segregation between the sexes.
197
P1: FYX
0521835224c11.xml
198
CUUK074B-Ruckstuhl
September 5, 2005
18:4
Richard Bon et al.
Social individuals
Let us consider now the case of a social species composed of subpopulations of females and males of size F and M with an intra-sex attraction
and no inter-sex attraction or repulsion. The probability of individual
animals moving from habitat A to B (B to A) is the same as earlier but,
in this case, as the number of same-sex individuals in the same habitat
increases (e.g. A), the probability of leaving this habitat decreases for
any individual (see Table 11.1; n = 1) as:
αF
(1 + F A )
and
βF
(1 + F B )
The proportions of females in habitats A and B are also independent
of what occurs for males, if there are no limits in space available (no
indirect competition).
At the equilibrium, the number of individual females f A moving
from A to B equals the number f B moving from B to A, so that:
βF F B
αF F A
=
(1 + F A )
(1 + F B )
or:
α F F A (1 + F B ) = β F F B (1 + F A )
α F F A − β F F B + F A F B (α F − β F ) = 0
As F B = F − F A , we obtain:
α F F A − β F (F − F A ) + F A (F − F A )(α F − β F ) = 0
(α F − β F )F A2 − (α F + β F + F (α F − β F ))F A + β F F = 0
Dividing by F 2 , we obtain:
αF + βF
βF
+ (α F − β F ) f A +
=0
(α F − β F ) f A2 −
F
F
If α F = β F , or r f = 1 then f A = f B = 0.5 whatever the total population
F. If α F < β F or r f < 1:
f A = 0.5(D + D 2 − 4E )
fB = 1 − fA
and
and if α F > β F or r f > 1:
f A = 0.5(D − D 2 − 4E )
and
fB = 1 − fA
with:
D =1+
rF + 1
(r F − 1)F
E =
1
(r F − 1)F
P1: FYX
0521835224c11.xml
CUUK074B-Ruckstuhl
September 5, 2005
18:4
From individual mechanisms to collective patterns
Taking into account the intra-sex attraction and r F = 1, the proportion
of females f A and f B will evolve as a function of the total subpopulation
F. For instance, if we set r F = 0.86 (as in the case of solitary animals), that
is to say a small preference of females for habitat A, we can see that
f A > 0.54 and grows with F (Fig. 11.4), which can be assimilated as
an effect of amplification. For large value of F, all the females will
be in habitat A ( f A ≈ 1) and environment B is neglected. In contrast,
if r F > 1, we will observe an amplification of the preference for the
environment B.
If the males have the same preference for habitats A and B, and
the same degree of attraction among each other as females, the habitat
segregation is nil. Besides, if males have a small and steady individual
preference for habitat B (β M = α F ), then the proportion of males in
habitat B increases with the number of males found in this habitat
and the degree of habitat segregation between the sexes will increase
with the increasing size of female and male subpopulations respectively
in habitats A and B. The curves of f A and f B are symmetrical if F = M.
This model shows how habitat segregation may be amplified by social
interactions and the population size of females and males, despite no
modification of individual habitat choice (r F , and r M constant).
It is also possible to show that the disequilibrium of proportions
of females in habitats A and B can be theoretically obtained without
initial differences of habitat preference, i.e when r F = 1. This may occur
if the probability for individual animals to move from A to B (B to A)
are very sensitive to the number of individuals in the habitat A (B) and
for example decreases with the square of the population (see Table 11.1,
n = 2):
αF
1+
F A2
and
βF
1 + F B2
In this case, most of the individuals are in habitat A ( f A ≈ 1, f B ≈ 0) or
in environment B ( f A ≈ 0, f B ≈ 1). The selection of the habitat A or B
is a random process and each environment has an equal probability to
be selected.
199